RESEARCH AND SUPPORTING MATERIAL

Every speech makes a quiet promise: that the person standing up knows something the room does not, and knows it well enough to be trusted with their time. Research is how you keep that promise. It is also, for most students, the part of speech preparation that feels the least like a skill and the most like luck — hours of scrolling, forty open tabs, and one usable sentence. It does not have to go that way. Research is a craft with repeatable moves: turning a thesis into questions a database can answer, searching with operators instead of hope, checking a source in ninety seconds by leaving it, and compressing a dense paragraph of findings into something an audience can absorb at speaking speed. This chapter also covers the step capable students most often skip — saying the source out loud so the credibility travels with the claim, and taking notes in a way that makes accidental plagiarism nearly impossible.

Section 1From Thesis to Search Plan

Take inventory before you search

The instinct is to open a search engine and start typing the topic. That is why research nights run long. A search engine will happily hand you two million results for a topic and zero results for the specific thing you needed, and you will not know which one you got until you have read for an hour. The cheaper move takes ten minutes and no internet: three columns on a page, headed what I already know, what I am assuming, and what an audience will make me prove.

Work it with a real thesis. Say you finished Chapter 6 with this: Our medical-surgical unit should adopt a structured bedside handoff. Under what I know you might write: I have watched the seven o’clock report happen in a hallway; I can describe the noise, the interruptions, the two nurses talking over a computer screen. Under what I am assuming: that unstructured handoffs cause most of the errors on the unit; that a structured tool would fix it; that the staff would tolerate the change. Under what I must prove: that communication failures during transitions actually contribute to patient harm; that a structured handoff format has been tested somewhere other than my imagination; that the added time at the bedside is small enough to be feasible; that “structured handoff” has an agreed definition rather than meaning whatever each unit decides it means.

That third column is your search plan, and notice how much more useful it is than the phrase “nursing communication.” Each line names a claim, and each claim implies a different kind of source. An incidence claim needs data from an agency or a study. A feasibility claim might be best answered by a fifteen-minute interview with a charge nurse. A definition claim needs a professional body, not a blog. You are no longer looking for material about a topic; you are looking for answers to four specific questions, which is a job with an end.

One more filter belongs here, and it comes straight from Chapter 5. Not every claim needs proof. Audiences grant some statements for free — that hospitals are busy, that sleep matters, that tuition has gone up. Spending a source on a claim nobody disputes is wasted airtime. Spending no source on a claim your audience actively doubts is a failed speech. Go down your list and mark each item free, cheap or expensive. That ranking is your evidence budget, and it tells you where the research hours should go.

Write questions a database can answer

A topic is a noun phrase. A research question is a sentence with a question mark and, crucially, boundaries. Databases are literal. They cannot infer that by “telehealth” you meant follow-up visits for rural adults with diabetes over the last five years. If you do not supply the boundaries, the database supplies sixty thousand results and lets you find out the hard way.

Health students will recognize the structure from PICO — population, intervention, comparison, outcome. You do not need the acronym, and the pattern works for any topic. Ask yourself four things: Who exactly are we talking about? Compared with what? Measured how? Over what period? A question missing the comparison is the single most common failure, because “does X work?” is unanswerable until you say “better than what?”

Look at what changed. Every stronger version could be typed into a database and produce a page of results you could reasonably skim. Every one of them also tells you when to stop researching, which is the part nobody mentions: you are done with a question when three credible sources converge on the same answer, or when you can honestly say they disagree and describe how.

Build a keyword ladder before you touch a search box

Your research question is written in your vocabulary. The literature is written in somebody else’s. British authors write “handover” where Americans write “handoff.” A researcher writes “care transition” where a nurse says “report.” If you search only your own words, you will find only the studies written by people who talk like you, and you will conclude that the research is thin when it is merely filed elsewhere.

So spend five minutes building a keyword ladder — a written list of the words the field might use for each concept in your question. Here is one for the bedside handoff question:

• Core concepts: handoff · structure · patient safety
• Synonyms and variants: handover, hand-off, shift report, sign-out, bedside report, change-of-shift report
• Narrower terms: SBAR, I-PASS, ISBAR, bedside shift report, nurse-to-nurse handoff
• Broader terms: care transitions, continuity of care, clinical communication, teamwork
• Outcome words: adverse event, sentinel event, medication error, near miss, patient satisfaction, length of stay
• Words to watch: “handoff” also appears in football, manufacturing and software — add a discipline word if the results wander

Two practical notes. First, harvest terms as you go: when a good article uses a phrase you did not have, add it to the ladder and rerun the search. Second, the abstract of any relevant article usually lists its own keywords or subject terms — those are free, professionally assigned vocabulary, and they are frequently better than anything you would have guessed.

Section 2Where the Good Material Lives

The source menu

Sources are not interchangeable. Each type answers a different question well and a different question badly, and a great deal of weak research is really just a mismatch — a trade magazine used for a statistical claim, or a thirty-page systematic review used for a human detail it was never going to contain. Before the table, one distinction worth having in your head. A primary source is the original record: the study itself, the interview transcript, the census data, the eyewitness account. A secondary source analyzes or reports on primary material: a review article, a news story about a study, a scholarly commentary. A tertiary source compiles and summarizes what secondary sources have already established: encyclopedias, most textbooks, subject guides, the summary paragraph at the top of a search result. Tertiary sources are excellent for orientation and poor as authority.

Search like a librarian

Your tuition already bought access to databases that a public search engine cannot reach. Use them. Google Scholar is a reasonable starting map — it is broad and it shows citation counts — but it mixes peer-reviewed articles with preprints, conference abstracts, class handouts and the occasional predatory journal, and it gives you no way to filter by study design or subject heading. A library database does. If your library carries CINAHL, Academic Search or PsycINFO, start there; PubMed is free to everyone through the National Library of Medicine and indexes the biomedical literature far more precisely than Scholar can. Use Scholar to chase what you find.

The operators below work in nearly every database and in most search engines. They are the difference between reading results and reading a haystack.

Put them together and a weak search becomes a working one. Weak: nursing handoff errors. Stronger: (handoff OR handover OR “shift report” OR “bedside report”) AND (standardiz* OR structured OR SBAR OR “I-PASS”) AND (error* OR “adverse event” OR “adverse events” OR “patient safety”), filtered to peer-reviewed articles from 2016 onward. Note the last group: most databases will not apply truncation inside a phrase in quotation marks, so plural forms of a phrase have to be listed separately. Then read the result count like a gauge. Four results means you over-constrained — drop the narrowest concept or widen a date range. Forty thousand means you under-constrained — add a third concept or move your terms into the title field. Somewhere between about twenty and two hundred is a page you can actually work.

Two techniques double the yield of any search. The first is citation chaining. When you find one genuinely good recent article, mine its reference list for everything it was built on — that is backward chaining — then use the “cited by” link in Google Scholar to find newer work that built on it, which is forward chaining. One strong review article plus twenty minutes of chaining routinely beats two hours of fresh searching. The second technique is asking a librarian. Every academic library staffs people whose entire job is this, most offer chat or fifteen-minute appointments, and students consistently underuse them. A librarian will hand you the right database and two subject headings in five minutes.

Primary research you can do yourself

Some of the best material in a student speech never appears in a database, because you generated it. An expert interview is the most underused tool in the course. Email a nurse manager, a physical therapist, a city planner, a shop steward — whoever actually does the thing you are speaking about — and ask for fifteen minutes. Say who you are, what the assignment is, how long it will take and what you want to ask. Most professionals say yes; being asked about their expertise is flattering.

Prepare six to eight open questions and put the most important one third, not first. Ask for stories and numbers, not opinions: “Walk me through what a handoff looks like on your unit at seven a.m.” produces better speech material than “Do you think handoffs are a problem?” Ask permission before recording. Read any sentence back to your source if you intend to quote it directly — misquoting a real person in front of a room is a serious error, and the fix costs one email.

You can also survey your own audience, which connects directly to the audience analysis work in Chapter 5. A five-question poll of your class gives you an opening hook that is undeniably about the people in the room. Be honest about what it is: twenty-two classmates are not a representative sample of anything, and saying so out loud actually raises your credibility rather than lowering it. “This is not a scientific sample — but of the twenty-two of you who answered on Tuesday, fourteen said you had worked a shift longer than twelve hours in the past month” is a strong, honest sentence.

Section 3Deciding What to Trust

Read laterally, not vertically

Most students were taught to evaluate a source by studying it: does the site look professional, is there an author, does it list references, does the address end in .org? Researchers at Stanford put that method to the test by watching how different groups evaluated unfamiliar websites. Professional fact-checkers reached accurate judgments far faster than either undergraduates or, notably, PhD historians — and they did it by doing almost the opposite of close reading. Within seconds they left the page and opened new tabs to find out what the rest of the internet said about the organization. That habit is called lateral reading, and the ordinary method — staying on the page and reading harder — is vertical reading. Vertical reading fails because every surface feature it examines is something a motivated publisher can fake in an afternoon.

The most portable version of lateral reading is Mike Caulfield’s SIFT framework: Stop, Investigate the source, Find better coverage, Trace claims to the original. It takes about ninety seconds and it is the single highest-return skill in this chapter.

You may also meet the CRAAP test — currency, relevance, authority, accuracy, purpose — developed at California State University, Chico, and still widely taught. It is a useful vocabulary for naming what you are checking. Its limitation is real and worth stating plainly: checklists focused on features visible on the page reward sites that are well designed, and researchers who have compared the approaches generally find lateral reading more accurate and faster. Opinion in the field genuinely differs about whether checklists still help beginners build the habit of asking the questions at all. The practical resolution is simple — use CRAAP’s questions, but answer them from other tabs.

What peer review does and does not guarantee

Peer review is the process in which a journal sends a submitted manuscript to two or three researchers in the same field, who recommend acceptance, revision or rejection. It is a meaningful filter. It catches methodological errors, unsupported conclusions, missing controls and overclaiming, and an article that has survived it has been read critically by people qualified to object.

It does not guarantee that a finding is true. Reviewers work from the manuscript, not the raw data; they rarely detect fabrication; and they cannot tell you whether a result will replicate. Papers are retracted every year — the Retraction Watch database exists precisely to track them — and some retracted work continues to be cited for a decade. There is also active scholarly disagreement about how well peer review performs as quality control at all: some researchers argue it is slow, inconsistent and conservative, and favor open or post-publication review; others hold that the alternatives have not proven better. Treat “peer reviewed” as a floor, not a ceiling.

Two related hazards deserve names. A predatory journal charges authors a fee and publishes nearly anything with only cosmetic review; such journals mimic legitimate titles closely enough to fool a search result. A useful screen: is the journal indexed in the databases your library carries, or listed in the Directory of Open Access Journals? Does the editorial board consist of findable people at real institutions? Does the website promise publication within days? The best-known blacklist of such publishers, maintained by the librarian Jeffrey Beall, was taken down in 2017, and no single free, authoritative list has replaced it, so the checking falls to you. A preprint, by contrast, is not deceptive at all — it is a legitimate manuscript posted before review. Preprints are fine to use. Just say what they are: “in a preprint posted in March, which has not yet been peer reviewed …”

Does the study actually support your sentence?

Suppose you have a real, peer-reviewed article in front of you. The last question is whether it supports the specific sentence you plan to say. Start with design. For questions about whether a treatment works, evidence is conventionally ranked with systematic reviews and meta-analyses at the top, then randomized controlled trials, then cohort studies, then case-control studies, then cross-sectional surveys, then case reports and expert opinion. That hierarchy is genuinely useful, and it is also frequently over-applied: it was built for questions about treatment effects, and it ranks poorly for questions about how people experience something, why a policy failed, or what a population believes — where well-conducted qualitative and survey research is the appropriate design, not a weaker one. Match the design to the question rather than reaching reflexively for the top of the ladder.

Then check the sample. Who was studied, how many, where, and when? A finding from 108 ICU nurses in one Midwestern hospital system in 2013 may still be worth citing — but you have to say those words, because the gap between that sample and “nurses” is where careless speakers overclaim. Check for conflicts of interest too; funding sources are disclosed at the end of most articles for exactly this reason, and a study of a product funded by its manufacturer is not disqualified but is worth mentioning.

Finally, watch the two words that swallow most research errors: cause and percent. Observational studies show association, not causation, and the phrase you want is usually “was associated with.” A worked example makes this concrete and honest. When Haynes and colleagues reported in the New England Journal of Medicine in 2009 on the World Health Organization’s nineteen-item surgical safety checklist, inpatient death rates at eight participating hospitals fell from 1.5 percent to 0.8 percent and complications from about 11 percent to about 7 percent. Those numbers are striking. They also came from a before-and-after study across eight hospitals rather than a randomized trial, so the responsible sentence is “death rates fell by roughly half after the checklist was introduced,” not “the checklist cut deaths in half.”

The percent problem is relative risk reduction versus absolute risk. “Cuts your risk in half” sounds enormous and tells you nothing, because a fifty percent relative reduction could mean falling from two cases in a hundred to one, or from forty in a hundred to twenty. Those are wildly different facts. Whenever a source gives you a percentage change, hunt for the underlying counts and give your audience both: “risk dropped by about half — from roughly two people in a hundred to one.” That single habit will make you more accurate than most of the sources you read.

Section 4Turning Sources into Supporting Material

The support toolbox

Supporting material is anything you offer to make a claim believable, understandable or memorable. Notice that those are three different jobs. A statistic proves; an example clarifies; a story makes the audience care. A speech that leans entirely on one form — usually statistics, because they feel most like research — will be simultaneously well documented and forgettable. The strongest main points typically stack two or three forms: a number for scale, an example for shape, and testimony for authority. The model sentences below are illustrative templates written for this table — copy the pattern, not the people or the numbers.

One caution about analogies, because they are the most abused tool on this list. A figurative analogy is an explanatory device, not evidence. It can make a complicated mechanism click in five seconds, and it can also smuggle in an argument the audience never agreed to. If your only support for a claim is a comparison, you have illustrated the claim rather than supported it — a distinction Chapter 13 returns to in detail.

Making numbers land

Numbers on a page and numbers in the ear are different animals. A reader can go back; a listener cannot. Every decimal you keep, every unnecessary digit, every unexplained percentage is a moment where part of the room quietly checks out. Six habits fix nearly all of it: round hard, use one number per sentence, always supply the denominator, translate into a human scale, anchor with a familiar comparison, and put the source before the number so the audience knows how much weight to give it.

Here is the same finding said three ways. The study and its numbers are invented for this example, so watch the moves rather than the content.

Weak: “A 2019 study demonstrated a 47.3 percent relative reduction in the incidence of unit-acquired infection, with a p-value of less than 0.05 and a confidence interval of 0.41 to 0.68.” Nobody heard the end of that sentence. The audience is still parsing “47.3.”

Better: “A 2019 study found that infections dropped by almost half.” Clear, but hollow — half of what?

Strongest: “In a 2019 study published in the Journal of Hospital Infection, infections on the units that adopted the protocol dropped by almost half — from about eight patients a month down to four. On a thirty-bed unit, that is one family a week who goes home on schedule instead of staying another six days.” Same finding. Now it has a source, a scale, a denominator and a picture.

Two more rules for spoken numbers. Never read a statistic off a slide word for word; say it in plain speech and let the slide hold the precision, which is Chapter 12’s territory. And be honest about uncertainty rather than hiding it — “estimates range from about four to about nine percent, depending on how the studies counted” is a stronger sentence than a fake-precise single figure, and audiences trust speakers who show them the edges of what is known.

The support sandwich

Evidence dropped into a speech without framing does nothing. Students often assume a good quotation speaks for itself; it does not, because the audience does not yet know what they are supposed to notice about it. The reliable pattern has five beats: claim (what you want them to accept), citation (who says so and when), evidence (the actual finding, quote or example), translation (what it means in plain language), and link (how it connects back to the claim or forward to the next point). Some instructors call this the support sandwich; the name matters less than never letting evidence sit naked in the middle of a paragraph.

Watch a real difference. Weak version: “Handoffs are a problem. The Joint Commission has said communication is a factor in sentinel events. Also a study found checklists helped. So we should use a structured handoff.” Every piece of that is roughly true and none of it is doing any work. The citations are vague, the evidence is asserted rather than delivered, and there is no translation at all.

Stronger version: “The riskiest ninety seconds of a hospital shift is the moment one nurse hands a patient to another. [Claim.] The Joint Commission, the body that accredits U.S. hospitals, has for years identified communication failures as one of the leading contributing factors in sentinel events — the safety events serious enough that a hospital has to investigate what happened. [Citation and evidence.] In other words, the thing most likely to hurt a patient is not a missing drug or a broken machine. It is a sentence that never got said. [Translation.] Which is exactly why the structure of that ninety seconds is worth arguing about. [Link.]”

The stronger version is longer, and that is the point. Support takes time, and time spent making one piece of evidence fully land is almost always a better investment than a fourth statistic delivered at speed. When your speech runs over, cut a main point rather than the material that makes the surviving points believable.

Section 5Citing Out Loud and Staying Honest

The anatomy of an oral citation

An oral citation is the spoken credit you give a source inside the speech itself. It is not the reference page, and it is not optional — a written bibliography your audience never sees does nothing for your credibility while you are talking. A complete oral citation names three things: the source, its authority (why this source is worth hearing) and the date. All three land before the claim rather than after it, because the audience needs to know how much weight to give a sentence while they are hearing it, not afterward. The repaired sentences below are patterns to imitate; the studies, authors and dates inside them are invented for the example, and yours must be real.

A few practical extensions. Cite an interview by name and role: “When I spoke with Maria Delgado, the charge nurse on 4 West, last Thursday, she told me…” Cite your own survey honestly: “in an informal poll of this class on Tuesday…” Cite a book by author and year, not by chapter and page. And vary the verb — found, reported, argues, estimates, concluded, warns, documented — because six consecutive sentences beginning “according to” will put a room to sleep faster than a dull topic.

Notes that make plagiarism nearly impossible

Almost no student sets out to plagiarize. What actually happens is a note-taking failure at eleven at night: a phrase gets copied into a document without quotation marks, three days pass, and by the time it reaches the outline it looks like something you wrote. The fix is mechanical. Record the citation at the moment of capture, every time, before you write anything else in the note. And label every note with one of three tags: Q for a direct quotation, written inside quotation marks in your notes; P for a paraphrase you built after closing the source; M for your own idea. Three letters, and the ambiguity that causes most accidental plagiarism disappears.

The subtle failure has a name. Patchwriting, a term from writing researcher Rebecca Moore Howard, is what happens when you keep a source’s sentence structure and swap in synonyms. It usually reflects an unfamiliar vocabulary rather than dishonest intent — and it is still plagiarism, because the thinking and the shape of the sentence remain someone else’s. Compare four versions of the same material:

Original source: “Structured handoff protocols reduce the omission of critical clinical information during transitions of care.”

Patchwriting (not acceptable): “Organized handoff procedures decrease the leaving out of important clinical facts during care transitions.” Nearly every content word was swapped and the last phrase was flipped around; the sentence is still the source’s sentence.

Genuine paraphrase (acceptable, still cited): “Researchers found that when nurses follow a set format, fewer critical details get dropped — the structure itself does the remembering.” Different shape, different emphasis, and the idea is credited.

Quotation (acceptable, cited): “The authors put it precisely: structured protocols ‘reduce the omission of critical clinical information during transitions of care.’”

The reliable technique for real paraphrase is to close the source, wait a beat, and say the idea out loud as you would explain it to a coworker. If you cannot explain it without the article open, you do not understand it yet — which is useful information, and a signal to read again rather than to reword. As for what needs a citation at all: common knowledge does not, but common knowledge is narrower than students assume and depends on the audience. That the heart has four chambers is common knowledge. That a specific percentage of hospital readmissions are preventable is not. When in doubt, cite — a citation has never once damaged a speech.

The working bibliography and the speaking outline

Start a working bibliography the moment you open the first source, not the night before the speech. One entry per source, and record six fields: author, title, where it was published, date, where you found it (database or URL and the DOI if there is one), and — the field students skip and later regret — one sentence on why this source matters and which claim it supports. That last line turns a list of citations into a map of your argument. Some instructors will ask you to formalize this as an annotated bibliography, which is the same thing with a fuller paragraph of summary and evaluation for each entry.

Then carry the citations into the speaking outline itself. Do not write “(source here)” and plan to improvise. Write the oral citation out in full, in the exact words you intend to say, in the exact spot where it belongs — Chapter 8 shows where those lines sit inside the outline structure. Oral citations are the most fumbled sentences in student speeches precisely because they are the only sentences nobody rehearses. Say them out loud in practice until the journal name comes off your tongue smoothly.

Finally, keep the written reference page in whatever format your course requires — APA in most health and social science programs, MLA in most humanities courses. The formatting rules are lookup work, not memory work; use your library’s citation guide and check the details. What is not lookup work is the underlying obligation, and it is worth stating plainly: you are telling a room full of people that something is true. Where that came from is part of the message, not an accessory to it.

Words worth knowing